
pip install -r requirements.txt


# import libraries 
import pandas as pd
import numpy as np 
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px
import requests
import pandas_datareader as web
import yfinance as yf
import math
%matplotlib inline
import time 
from datetime import date
from sklearn.preprocessing import MinMaxScaler
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error
from keras.models import Sequential
from keras.layers.core import Dense, Activation, Dropout
from keras.layers import LSTM
from keras.callbacks import EarlyStopping
# py Portfolio Optimization 
from pypfopt.expected_returns import mean_historical_return
from pypfopt import risk_models,objective_functions
from pypfopt.efficient_frontier import EfficientFrontier
from pypfopt.discrete_allocation import DiscreteAllocation, get_latest_prices
import warnings
warnings.filterwarnings("ignore")


# get the tickers from nasdaq for mage and large companies
def get_tickers():
    df = pd.read_csv('nasdaq_screener.csv')['Symbol']
    return df.tolist()
all_tickers = get_tickers()


# Adjusted close for all tickers
def get_data(tickers):
  
    start_year = date.today().year-6
    start_date = str('{}-01-01'.format(start_year))
    end_date = date.today().strftime("%Y-%m-%d")

    df = yf.download(tickers, start_date, end_date) 
    return df['Adj Close']


df = get_data(all_tickers)

[*********************100%***********************]  193 of 193 completed


# fill nan value 
df.fillna(axis=1,method='ffill',inplace=True)
df.fillna(axis=1,method='bfill',inplace = True)


pandemic_data = df['2020-01-01':'2022-01-01']


# preprocessing data for visualizing 
BCES = pd.read_csv('nasdaq_screener.csv')
BCES1=  pd.merge(BCES[['Sector','Symbol']],pandemic_data.T.reset_index().rename(columns={'index':'Symbol'}),how='right',on='Symbol')\
.set_index(['Sector','Symbol'])
BCES1


df1 = BCES1.groupby(['Sector']).mean().T
df1


px.line(df1, labels={"index":'Date',"value": "Adj Close Price"}
        ,title = "Adj close prices for mega and large INC. in each sector")


df1 = df1/df1.iloc[0,:] # for cumulative return
df1


px.line(df1, labels={"index":'Date',"value": "cumulative return"}
        ,title = "Mean Adj close prices for mega and large INC. in each sector Cumulated")


FANNG = pandemic_data[['META','AAPL','AMZN','NFLX','GOOGL']]
FANNG = FANNG / FANNG.iloc[0,:]


px.line(FANNG,labels={"value":'Cumulative Return',"variable": "ticker"}
        ,title = "FANNG stocks Cumulative Return")


FANNG = ['META','AAPL','AMZN','NFLX','GOOGL']
FAANG = get_data(FANNG)

[*********************100%***********************]  5 of 5 completed


def data_prep(data,ticker,time_step=60):
    ticker = ticker.upper()
    df = data.filter([ticker]).values # filter the data by ticker
    X= []
    y =[]
    for i in range(len(df)-time_step): # pre_processing
        X.append(df[i:time_step + i,0]) # features numbers = time_step
        y.append(df[i+time_step,0]) 

    # covnert to array
    X,y = np.array(X),np.array(y) 

    return X,y


def split_data(X,y,split_percentage=0.75):
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=1-split_percentage,shuffle=False)
    return X_train, X_test, y_train, y_test


def scale_data(X_train, X_test, y_train,y_test,time_step=60):
    scaler = MinMaxScaler(feature_range=(0,1))
    X_train = scaler.fit_transform(X_train)
    y_train = scaler.fit_transform(y_train.reshape(-1,1))
    X_test = scaler.fit_transform(X_test)
    y_test = scaler.fit_transform(y_test.reshape(-1,1))
    return X_train, X_test, y_train,y_test, scaler


def plot_result(data, ticker,predictions,training_shape,model,time_step=60):
    ticker = ticker.upper()
    data = data.filter([ticker])
    train = data[:training_shape+time_step+1]
    test = data[training_shape+time_step:]
    test['predictions'] = predictions
    plt.figure(figsize=(15,8))
    plt.plot(train[ticker])
    plt.plot(test[[ticker,'predictions']])
    plt.legend(['trian','Val','val_pred'])
    plt.xlabel('Date')
    plt.ylabel('Adj close Price')
    plt.title(f'{model} {ticker} Stock prediction');


def build_lr(df, ticker):
    X, y = data_prep(df,ticker)
    X_train, X_test, y_train, y_test = split_data(X,y)
    X_train, X_test, y_train,y_test, scaler = scale_data(X_train, X_test, y_train,y_test)
    LR = LinearRegression()
    tic = time.time()
    LR.fit(X_train, y_train)
    tac = time.time()
    predictions = LR.predict(X_test)
    predictions = scaler.inverse_transform(predictions.reshape(-1,1))
    y_test1 = scaler.inverse_transform(y_test)
    MSE = mean_squared_error(y_test1, predictions)
    print('MSE : ', MSE)
    print(f'LR Model Take {math.ceil(tac-tic)} Sec')
    #print(X_train.shape,y_train.shape,X_test.shape,y_test.shape)
    return predictions, X_train.shape[0], LR,X_test,y_test,scaler


def build_LSTM(df, ticker):
    X, y = data_prep(df,ticker)
    X_train, X_test, y_train, y_test = split_data(X,y)
    X_train, X_test, y_train,y_test, scaler = scale_data(X_train, X_test, y_train,y_test)
    # reshape the data to be 3D for LSTM model
    X_train = np.reshape(X_train, (X_train.shape[0],X_train.shape[1],1))
    X_test = np.reshape(X_test , (X_test.shape[0],X_test.shape[1],1))

  #Build LSTM model
    model=Sequential()
    model.add(LSTM(128,return_sequences=True,input_shape=(X_train.shape[1],1)))
    model.add(LSTM(64))
    model.add(Dense(50))
    model.add(Dense(1))
    # compile the model and train
    model.compile(optimizer='adam', loss = 'mean_squared_error')
    #earlyStop=EarlyStopping(monitor="val_loss",verbose=2,mode='min',patience=5)
    #model.fit(X_train,y_train,epochs=100,batch_size=32,validation_data=(X_test,y_test),callbacks=[earlyStop])
    print('Training....')
    tic = time.time()
    model.fit(X_train,y_train,epochs=20,batch_size=8,verbose=0)
    tac = time.time()
    predictions = model.predict(X_test)
    predictions = scaler.inverse_transform(predictions.reshape(-1,1))
    y_test1 = scaler.inverse_transform(y_test)
    MSE = mean_squared_error(y_test1, predictions)
    print('MSE : ', MSE)
    print(f'LSTM Model Take {math.ceil(tac-tic)} Sec')
    return predictions, X_train.shape[0],LSTM,X_test,y_test,scaler


predictions,split_index,LR,X_test,y_test,scaler = build_lr(FAANG,'meta')
plot_result(FAANG,'meta',predictions ,split_index,'LR')

MSE :  61.69505302036034
LR Model Take 1 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = build_LSTM(FAANG,'meta')
plot_result(FAANG,'meta',predictions ,split_index,'LSTM')

Training....
14/14 [==============================] - 1s 41ms/step
MSE :  57.16863528743344
LSTM Model Take 174 Sec


predictions,split_index,LR,X_test,y_test,scaler = build_lr(FAANG,'aapl')
plot_result(FAANG,'aapl',predictions ,split_index,'LR')

MSE :  9.240025707760518
LR Model Take 1 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = build_LSTM(FAANG,'aapl')
plot_result(FAANG,'aapl',predictions ,split_index,'LSTM')

Training....
14/14 [==============================] - 2s 38ms/step
MSE :  9.532977231229212
LSTM Model Take 205 Sec


predictions,split_index,LR,X_test,y_test,scaler = build_lr(FAANG,'amzn')
plot_result(FAANG,'amzn',predictions ,split_index,'LR')

MSE :  13.953507909063525
LR Model Take 1 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = build_LSTM(FAANG,'amzn')
plot_result(FAANG,'amzn',predictions ,split_index,'LSTM')

Training....
14/14 [==============================] - 2s 53ms/step
MSE :  15.369023641495627
LSTM Model Take 175 Sec


predictions,split_index,LR,X_test,y_test,scaler = build_lr(FAANG,'nflx')
plot_result(FAANG,'nflx',predictions ,split_index,'LR')

MSE :  182.82522576074552
LR Model Take 1 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = build_LSTM(FAANG,'nflx')
plot_result(FAANG,'nflx',predictions ,split_index,'LSTM')

Training....
14/14 [==============================] - 1s 38ms/step
MSE :  220.91215387702138
LSTM Model Take 184 Sec


predictions,split_index,LR,X_test,y_test,scaler = build_lr(FAANG,'googl')
plot_result(FAANG,'googl',predictions ,split_index,'LR')

MSE :  6.658735372888752
LR Model Take 1 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = build_LSTM(FAANG,'googl')
plot_result(FAANG,'googl',predictions ,split_index,'LSTM')

Training....
14/14 [==============================] - 1s 39ms/step
MSE :  6.125015813108856
LSTM Model Take 172 Sec


def refine_LSTM1(df, ticker):
    X, y = data_prep(df,ticker)
    X_train, X_test, y_train, y_test = split_data(X,y)
    X_train, X_test, y_train,y_test, scaler = scale_data(X_train, X_test, y_train,y_test)
    # reshape the data to be 3D for LSTM model
    X_train = np.reshape(X_train, (X_train.shape[0],X_train.shape[1],1))
    X_test = np.reshape(X_test , (X_test.shape[0],X_test.shape[1],1))

    #Build LSTM model
    model=Sequential()
    model.add(LSTM(128,return_sequences=True,input_shape=(X_train.shape[1],1)))
    model.add(LSTM(64))
    model.add(Dense(50))
    model.add(Dense(1))
    # compile the model and train
    model.compile(optimizer='adam', loss = 'mean_squared_error')
    earlyStop=EarlyStopping(monitor="val_loss",verbose=0,mode='min',patience=5)
    print('Training....')
    tic = time.time()
    model.fit(X_train,y_train,epochs=15,batch_size=4,validation_data=(X_test,y_test),callbacks=[earlyStop],verbose=0)

    #model.fit(X_train,y_train,epochs=20,batch_size=8,verbose=0)
    tac = time.time()
    predictions = model.predict(X_test)
    predictions = scaler.inverse_transform(predictions.reshape(-1,1))
    y_test1 = scaler.inverse_transform(y_test)
    MSE = mean_squared_error(y_test1, predictions)
    print('MSE : ', MSE)
    print(f'LSTM Model Take {math.ceil(tac-tic)} Sec')
    return predictions, X_train.shape[0],LSTM,X_test,y_test,scaler


predictions,split_index,LSTM,X_test,y_test,scaler = refine_LSTM1(FAANG,'meta')

Training....
14/14 [==============================] - 2s 46ms/step
MSE :  58.174018303136585
LSTM Model Take 329 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = refine_LSTM1(FAANG,'aapl')

Training....
14/14 [==============================] - 2s 44ms/step
MSE :  9.206769120173036
LSTM Model Take 299 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = refine_LSTM1(FAANG,'amzn')

Training....
14/14 [==============================] - 2s 50ms/step
MSE :  13.148163923526266
LSTM Model Take 302 Sec


def refine_LSTM2(df, ticker):
    X, y = data_prep(df,ticker)
    X_train, X_test, y_train, y_test = split_data(X,y)
    X_train, X_test, y_train,y_test, scaler = scale_data(X_train, X_test, y_train,y_test)
    # reshape the data to be 3D for LSTM model
    X_train = np.reshape(X_train, (X_train.shape[0],X_train.shape[1],1))
    X_test = np.reshape(X_test , (X_test.shape[0],X_test.shape[1],1))

    #Build LSTM model
    model=Sequential()
    model.add(LSTM(120,return_sequences=True,input_shape=(X_train.shape[1],1)))
    model.add(LSTM(60))
    model.add(Dense(50))
    model.add(Dense(1))
    # compile the model and train
    model.compile(optimizer='adam', loss = 'mean_squared_error')
    #earlyStop=EarlyStopping(monitor="val_loss",verbose=0,mode='min',patience=5)
    print('Training....')
    tic = time.time()
    #model.fit(X_train,y_train,epochs=50,batch_size=32,validation_data=(X_test,y_test),callbacks=[earlyStop],verbose=0)

    model.fit(X_train,y_train,epochs=50,batch_size=32,verbose=0)
    tac = time.time()
    predictions = model.predict(X_test)
    predictions = scaler.inverse_transform(predictions.reshape(-1,1))
    y_test1 = scaler.inverse_transform(y_test)
    MSE = mean_squared_error(y_test1, predictions)
    print('MSE : ', MSE)
    print(f'LSTM Model Take {math.ceil(tac-tic)} Sec')
    return predictions, X_train.shape[0],LSTM,X_test,y_test,scaler


predictions,split_index,LSTM,X_test,y_test,scaler = refine_LSTM2(FAANG,'meta')

Training....
14/14 [==============================] - 1s 25ms/step
MSE :  58.44141211188549
LSTM Model Take 149 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = refine_LSTM2(FAANG,'aapl')

Training....
14/14 [==============================] - 1s 24ms/step
MSE :  10.31142097849755
LSTM Model Take 148 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = refine_LSTM2(FAANG,'amzn')

Training....
14/14 [==============================] - 1s 25ms/step
MSE :  15.063305170647936
LSTM Model Take 145 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = refine_LSTM2(FAANG,'nflx')

Training....
14/14 [==============================] - 1s 23ms/step
MSE :  217.7250558179343
LSTM Model Take 144 Sec


predictions,split_index,LSTM,X_test,y_test,scaler = refine_LSTM2(FAANG,'GOOGL')

Training....
14/14 [==============================] - 1s 23ms/step
MSE :  7.454766115854251
LSTM Model Take 147 Sec


predictions,split_index,LR,X_test,y_test,scaler = build_lr(df,'ZBRA')
plot_result(df,'ZBRA',predictions ,split_index,'LR')

MSE :  98.98113464222868
LR Model Take 1 Sec


def prediction_future(X_test,y_test,time_step=60):
    temp1 = X_test[-1][1:].tolist()
    predictions = y_test[-1].tolist()
    temp1.extend(predictions)
  
    for i in range(30):
        if len(temp1)==time_step:
            temp1 = np.array(temp1).reshape(-1,time_step)
            predic = LR.predict(temp1)
            temp1 = temp1.tolist()
            predic = predic[0].tolist()
            temp1[0].extend(predic)
            predictions.extend(predic)
            print('done')
        else:
            temp1 = temp1[0][1:]
            temp1 = np.array(temp1).reshape(-1,time_step)
            predic = LR.predict(temp1)
            temp1 = temp1.tolist()
            predic = predic[0].tolist()
            temp1[0].extend(predic)
            predictions.extend(predic)
    return predictions


predictions = prediction_future(X_test,y_test)

done


predictions = scaler.inverse_transform(np.array(predictions).reshape(-1,1))


def show_result(predictions,ticker):
    week = float(predictions[7] / predictions[0] *100-100)
    two_week = float(predictions[14] / predictions[0] *100-100)
    month = float(predictions[-1] / predictions[0] *100-100)
    print(f'regarding to price today {round(float(predictions[0]),2)}$ for {ticker}')
    print(f'predicted stock value for 7 days  is  : {round(week,2)} %')
    print(f'predicted stock value for 14 days is  : {round(two_week,2)} %')
    print(f'predicted stock value for 30 days is  : {round(month,2)} %')


show_result(predictions,'ZBRA')

regarding to price today 248.22$ for ZBRA
predicted stock value for 7 days  is  : 0.61 %
predicted stock value for 14 days is  : -0.57 %
predicted stock value for 30 days is  : -2.27 %


df = get_data(all_tickers)

[*********************100%***********************]  193 of 193 completed


data = df[:'2018-12-31']# filter data
# Drop nan value 
data.dropna(axis=1,inplace = True)


def port_opt(df):
    intention = int(input("input\n1 for maximun sharpe ratio\n2 for minimun risk\n3 for maximum return\n"))
    total_value = int(input('How much would you invest $'))
    print()
    mu = mean_historical_return(df)
    S = risk_models.sample_cov(df)
    ef = EfficientFrontier(mu, S)
    ef.add_objective(objective_functions.L2_reg, gamma=1)
    if intention == 1:
        ef.max_sharpe()
    elif intention == 2 :
        ef.min_volatility()
    else:
        ef._max_return()
    cleaned_weights = ef.clean_weights()
    ef.portfolio_performance(verbose=True)
    print()
    latest_prices = get_latest_prices(df)
    da = DiscreteAllocation(cleaned_weights, latest_prices, total_portfolio_value=total_value)
    allocation, leftover = da.greedy_portfolio()
    name = pd.read_csv('nasdaq_screener.csv')[['Symbol','Name']]
    name.columns = ['Ticker','Name']
    alloc = pd.DataFrame(allocation.items(),columns=['Ticker','Numbers_of_shares'])
    alloc['Allocation'] = round(alloc['Numbers_of_shares'] / alloc['Numbers_of_shares'].sum(),4)*100
    alloc = pd.merge(alloc,name,how='left',on='Ticker')
  
    if len(alloc) > 0:
        print('Orderd Allocation\n',alloc.to_string())
        print('Left over : $',int(leftover))

    else:
        print('Low money')
        port_opt(df)


port_opt(data)

input
1 for maximun sharpe ratio
2 for minimun risk
3 for maximum return
1
How much would you invest $2000

Expected annual return: 40.4%
Annual volatility: 18.0%
Sharpe Ratio: 2.14

Orderd Allocation
    Ticker  Numbers_of_shares  Allocation                                             Name
0    ETSY                  2        6.67                           Etsy Inc. Common Stock
1     AMD                  6       20.00         Advanced Micro Devices Inc. Common Stock
2    ABMD                  1        3.33                        ABIOMED Inc. Common Stock
3     CZR                  2        6.67          Caesars Entertainment Inc. Common Stock
4    NVDA                  2        6.67                  NVIDIA Corporation Common Stock
5     CME                  1        3.33              CME Group Inc. Class A Common Stock
6    MTCH                  1        3.33                    Match Group Inc. Common Stock
7    TTWO                  1        3.33  Take-Two Interactive Software Inc. Common Stock
8    ALGN                  1        3.33               Align Technology Inc. Common Stock
9     CSX                  2        6.67                     CSX Corporation Common Stock
10   CPRT                  2        6.67                    Copart Inc. (DE) Common Stock
11   ISRG                  1        3.33             Intuitive Surgical Inc. Common Stock
12    KDP                  2        6.67               Keurig Dr Pepper Inc. Common Stock
13   IDXX                  1        3.33             IDEXX Laboratories Inc. Common Stock
14   AXON                  1        3.33                Axon Enterprise Inc. Common Stock
15   MKTX                  1        3.33           MarketAxess Holdings Inc. Common Stock
16    EXC                  1        3.33                  Exelon Corporation Common Stock
17   LOGI                  1        3.33      Logitech International S.A. Ordinary Shares
18   FTNT                  1        3.33                       Fortinet Inc. Common Stock
Left over : $ 5


port_opt(data)

input
1 for maximun sharpe ratio
2 for minimun risk
3 for maximum return
2
How much would you invest $2000

Expected annual return: 12.5%
Annual volatility: 12.2%
Sharpe Ratio: 0.86

Orderd Allocation
    Ticker  Numbers_of_shares  Allocation                                               Name
0     XEL                  1        3.33                      Xcel Energy Inc. Common Stock
1     AEP                  1        3.33  American Electric Power Company Inc. Common Stock
2     LNT                  1        3.33            Alliant Energy Corporation Common Stock
3     KDP                  1        3.33                 Keurig Dr Pepper Inc. Common Stock
4     EXC                  1        3.33                    Exelon Corporation Common Stock
5     PEP                  1        3.33                          PepsiCo Inc. Common Stock
6     REG                  1        3.33           Regency Centers Corporation Common Stock
7    GLPI                  1        3.33    Gaming and Leisure Properties Inc. Common Stock
8    ERIE                  1        3.33        Erie Indemnity Company Class A Common Stock
9     CME                  1        3.33                CME Group Inc. Class A Common Stock
10   MKTX                  1        3.33             MarketAxess Holdings Inc. Common Stock
11   CHRW                  1        3.33          C.H. Robinson Worldwide Inc. Common Stock
12   COST                  1        3.33          Costco Wholesale Corporation Common Stock
13   SBUX                  1        3.33                 Starbucks Corporation Common Stock
14   EQIX                  1        3.33                     Equinix Inc. Common Stock REIT
15    KHC                  1        3.33               The Kraft Heinz Company Common Stock
16   CINF                  1        3.33      Cincinnati Financial Corporation Common Stock
17   NDAQ                  1        3.33                           Nasdaq Inc. Common Stock
18   JKHY                  1        3.33          Jack Henry & Associates Inc. Common Stock
19   UHAL                  1        3.33                                Amerco Common Stock
20   LAMR                  1        3.33     Lamar Advertising Company Class A Common Stock
21   HSIC                  1        3.33                     Henry Schein Inc. Common Stock
22   SIRI                  3       10.00               Sirius XM Holdings Inc. Common Stock
23    GEN                  1        3.33                      Gen Digital Inc. Common Stock
24   HBAN                  1        3.33    Huntington Bancshares Incorporated Common Stock
25    NWS                  1        3.33              News Corporation Class B Common Stock
26   NWSA                  1        3.33              News Corporation Class A Common Stock
27   LSCC                  1        3.33     Lattice Semiconductor Corporation Common Stock
Left over : $ 5


port_opt(data)

input
1 for maximun sharpe ratio
2 for minimun risk
3 for maximum return
3
How much would you invest $2000

Expected annual return: 88.7%
Annual volatility: 69.8%
Sharpe Ratio: 1.24

Orderd Allocation
   Ticker  Numbers_of_shares  Allocation                                      Name
0    AMD                108       100.0  Advanced Micro Devices Inc. Common Stock
Left over : $ 6

		2020-01-02	2020-01-03	2020-01-06	2020-01-07	2020-01-08	2020-01-09	2020-01-10	2020-01-13	2020-01-14	2020-01-15	...	2021-12-17	2021-12-20	2021-12-21	2021-12-22	2021-12-23	2021-12-27	2021-12-28	2021-12-29	2021-12-30	2021-12-31
Sector	Symbol
Technology	AAPL	73.561546	72.846359	73.426811	73.081505	74.257118	75.834389	76.005829	77.629639	76.581383	76.253204	...	170.152466	168.770493	171.991791	174.626495	175.262802	179.289444	178.255447	178.344925	177.171738	176.545380
Health Care	ABMD	168.809998	166.820007	179.039993	180.350006	178.690002	183.600006	189.059998	168.100006	172.729996	177.919998	...	315.549988	316.230011	332.200012	343.399994	352.209991	357.829987	357.440002	361.839996	362.059998	359.170013
Consumer Discretionary	ABNB	168.809998	166.820007	179.039993	180.350006	178.690002	183.600006	189.059998	168.100006	172.729996	177.919998	...	157.910004	157.229996	165.660004	169.289993	171.309998	171.679993	169.710007	167.440002	168.779999	166.490005
Technology	ADBE	334.429993	331.809998	333.709991	333.390015	337.869995	340.450012	339.809998	345.630005	344.630005	342.940002	...	556.640015	549.770020	557.520020	563.979980	569.619995	577.679993	569.359985	569.289978	570.530029	567.059998
Technology	ADI	113.905487	111.900352	110.585663	113.101562	114.123039	114.123039	112.146255	112.609718	113.054253	111.143692	...	166.949066	165.192337	168.882477	169.088562	169.451675	172.209473	171.139740	172.258560	171.532288	172.503906
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
Utilities	XEL	57.911705	58.190212	58.106667	57.985973	57.930279	58.060242	58.153076	58.830788	58.923626	59.851997	...	65.183388	65.300140	64.984711	65.366753	65.249199	65.454910	66.238571	66.571617	66.836105	66.316925
Technology	ZBRA	259.140015	256.049988	258.010010	256.470001	247.639999	246.500000	246.270004	248.580002	248.070007	247.800003	...	588.520020	570.780029	580.210022	580.429993	582.409973	606.330017	597.320007	601.119995	597.539978	595.200012
	ZI	259.140015	256.049988	258.010010	256.470001	247.639999	246.500000	246.270004	248.580002	248.070007	247.800003	...	62.730000	61.320000	64.580002	64.900002	65.720001	66.820000	64.720001	64.489998	64.730003	64.199997
	ZM	68.720001	67.279999	70.320000	71.900002	72.550003	72.620003	73.089996	74.029999	73.160004	76.940002	...	199.740005	197.970001	199.419998	193.130005	191.649994	187.440002	183.770004	181.809998	190.520004	183.910004
	ZS	47.330002	47.380001	48.700001	48.400002	50.750000	54.000000	55.000000	59.619999	59.540001	59.990002	...	304.869995	305.940002	319.420013	322.010010	323.000000	329.890015	324.660004	323.179993	324.589996	321.329987

Sector	Basic Materials	Consumer Discretionary	Consumer Staples	Energy	Finance	Health Care	Industrials	Real Estate	Technology	Telecommunications	Utilities
2020-01-02	34.411755	150.009759	57.741906	39.725860	135.046585	147.254161	88.918388	161.905279	110.985927	70.424459	62.365625
2020-01-03	33.801716	149.395973	57.585885	40.071862	134.402880	146.086201	88.144113	163.403870	110.158006	70.415781	62.107422
2020-01-06	33.274864	149.475932	57.783164	40.141175	133.681833	147.908884	87.636535	163.240372	110.734639	70.652268	62.011530
2020-01-07	33.210163	149.849660	57.220908	42.011181	133.747865	147.654255	87.867168	162.007652	111.362158	70.806174	62.079539
2020-01-08	33.459721	150.514730	56.766306	41.336972	133.497770	148.942633	88.162891	162.926027	112.056666	70.908054	62.073946
...	...	...	...	...	...	...	...	...	...	...	...
2021-12-27	62.074051	205.875998	73.945442	53.746246	211.448817	226.758233	162.369727	241.019228	223.817161	91.164978	79.231415
2021-12-28	62.366943	205.453546	74.375756	53.521850	210.456668	225.649232	162.372701	241.798194	221.048634	91.419141	79.409990
2021-12-29	62.894154	205.907762	74.871976	53.013877	208.890593	227.965054	163.472283	242.726641	221.036981	91.380869	79.844394
2021-12-30	62.298603	205.951681	74.663324	52.124077	207.229121	227.991423	162.086330	244.383500	220.172799	91.160958	79.574097
2021-12-31	62.542675	205.661640	75.127001	52.273855	208.033916	226.388668	162.772853	244.439037	218.898556	90.515361	79.695093

Sector	Basic Materials	Consumer Discretionary	Consumer Staples	Energy	Finance	Health Care	Industrials	Real Estate	Technology	Telecommunications	Utilities
2020-01-02	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000
2020-01-03	0.982272	0.995908	0.997298	1.008710	0.995233	0.992068	0.991292	1.009256	0.992540	0.999877	0.995860
2020-01-06	0.966962	0.996441	1.000715	1.010455	0.989894	1.004446	0.985584	1.008246	0.997736	1.003235	0.994322
2020-01-07	0.965082	0.998933	0.990977	1.057527	0.990383	1.002717	0.988178	1.000632	1.003390	1.005420	0.995413
2020-01-08	0.972334	1.003366	0.983104	1.040556	0.988531	1.011466	0.991503	1.006305	1.009648	1.006867	0.995323
...	...	...	...	...	...	...	...	...	...	...	...
2021-12-27	1.803862	1.372417	1.280620	1.352928	1.565747	1.539911	1.826053	1.488643	2.016627	1.294507	1.270434
2021-12-28	1.812373	1.369601	1.288072	1.347280	1.558401	1.532379	1.826087	1.493455	1.991682	1.298116	1.273297
2021-12-29	1.827694	1.372629	1.296666	1.334493	1.546804	1.548106	1.838453	1.499189	1.991577	1.297573	1.280263
2021-12-30	1.810387	1.372922	1.293053	1.312094	1.534501	1.548285	1.822866	1.509423	1.983790	1.294450	1.275929
2021-12-31	1.817480	1.370988	1.301083	1.315865	1.540460	1.537401	1.830587	1.509766	1.972309	1.285283	1.277869

	META	AAPL	AMZN	NFLX	GOOGL	Time_Sec(Mean)
LR	62.54	9.24	13.95	182.83	6.66	`< 1`
LSTM	57.17	9.53	15.37	220.91	6.13	`182`
LSTM Refine 1	58.17	9.21	13.15	`nan`	`nan`	`286`
LSTM Refine 2	58.44	10.31	15.06	220.57	6.88	`148`

Stock price Predictor & Portfolio Optimization¶

How each sector perform in the pandemic¶

For Sectors¶

For FANNG stocks¶

Data Preprocessing¶

Training models¶

Try to Refine Lstm Model¶

Mean Squared Error By Each Model¶

Get the results for the future¶

Portfolio Optimization¶

maximun sharpe ratio¶

minimun risk¶

maximum return¶

Justification¶

Conclusion¶

Improvements¶

	Layer1	Layer2	Layer3	Layer4	epochs	Batch_size	Early_stop
LSTM	LSTM(128)	LSTM(64)	Dense(50)	Dense(1)	20	8	No
LSTM Refine1	LSTM(128)	LSTM(64)	Dense(50)	Dense(1)	15	4	yes
LSTM Refine2	LSTM(120)	LSTM(60)	Dense(50)	Dense(1)	50	32	No